In 3GPP LTE (3rd. Generation Partnership Project Long Term Evolution), a Zadoff-Chu sequence (“ZC sequence”) is adopted as a reference signal (“RS”) that is used in uplink. The reason for adopting a ZC sequence as an RS is that a ZC sequence has a uniform frequency characteristic and has good auto-correlation and cross-correlation characteristics. A ZC sequence is a kind of CAZAC (Constant Amplitude and Zero Auto-correlation Code) sequence and represented by following equation 1 or equation 2.
                    (                  Equation          ⁢                                          ⁢          1                )                                                                                  a            r                    ⁡                      (            k            )                          =                  {                                                                                          ⅇ                                                                  -                        j                                            ⁢                                                                        2                          ⁢                          π                          ⁢                                                                                                          ⁢                          r                                                N                                            ⁢                                              (                                                                                                            k                              2                                                        /                            2                                                    +                          qk                                                )                                                                              ,                                                                              N                  :                  even                                                                                                                          ⅇ                                                                  -                        j                                            ⁢                                                                        2                          ⁢                          π                          ⁢                                                                                                          ⁢                          r                                                N                                            ⁢                                              (                                                                                                            k                              ⁡                                                              (                                                                  k                                  +                                  1                                                                )                                                                                      /                            2                                                    +                          qk                                                )                                                                              ,                                                                              N                  :                  odd                                                                                        [        1        ]                                (                  Equation          ⁢                                          ⁢          2                )                                                                                  a            r                    ⁡                      (            k            )                          =                  {                                                                                          ⅇ                                          j                      ⁢                                                                        2                          ⁢                          π                          ⁢                                                                                                          ⁢                          r                                                N                                            ⁢                                              (                                                                                                            k                              2                                                        /                            2                                                    +                          qk                                                )                                                                              ,                                                                              N                  :                  even                                                                                                                          ⅇ                                          j                      ⁢                                                                        2                          ⁢                          π                          ⁢                                                                                                          ⁢                          r                                                N                                            ⁢                                              (                                                                                                            k                              ⁡                                                              (                                                                  k                                  +                                  1                                                                )                                                                                      /                            2                                                    +                          qk                                                )                                                                              ,                                                                              N                  :                  odd                                                                                                    
In equation 1 and equation 2, “N” is the sequence length, “r” is the ZC sequence number, and “N” and “r” are coprime. Also, “q” is an arbitrary integer. It is possible to generate N−1 quasi-orthogonal sequences of good cross-correlation characteristics from a ZC sequence having the sequence length N of a prime number. In this case, the cross-correlation is constant at √N between the N−1 quasi-orthogonal sequences generated.
Here, in the RS's that are used in uplink, the reference signal for channel estimation used to demodulate data (i.e. DM-RS (Demodulation Reference Signal)) is transmitted in the same band as the data transmission bandwidth. That is, when the data transmission bandwidth is narrow, a DM-RS is also transmitted in a narrow band. For example, if the data transmission bandwidth is one RB (Resource Block), the DM-RS transmission bandwidth is also one RB. Likewise, if the data transmission bandwidth is two RB's, the DM-RS transmission bandwidth is also two RB's. Also, in 3GPP LTE, one RB is comprised of twelve subcarriers. Consequently, a ZC sequence having a sequence length N of 11 or 13 is used as a DM-RS that is transmitted in one RB, and a ZC sequence having a sequence length N of 23 or 29 is used as a DM-RS that is transmitted in two RB's. Here, when a ZC sequence having a sequence length N of 11 or 23 is used, a DM-RS of 12 subcarriers or 24 subcarriers is generated by cyclically expanding the sequence, that is, by copying the head data of the sequence to the tail end of the sequence. On the other hand, when a ZC sequence having a sequence length N of 13 or 29 is used, a DM-RS of 12 subcarriers or 24 subcarriers is generated by performing truncation, that is, by deleting part of the sequence.
As a method of allocating ZC sequences, to reduce the interference between DM-RS's that are used between different cells, that is, to reduce the inter-cell interference of DM-RS, in each RB, ZC sequences of different sequence numbers are allocated to adjacent cells as DM-RS's. The data transmission bandwidth is determined by the scheduling in each cell, and therefore DM-RS's of different transmission bandwidths are multiplexed between cells. However, if ZC sequences of different transmission bandwidths, that is, ZC sequences of different sequence lengths, are multiplexed, a specific combination of ZC sequence numbers has a high cross-correlation.
FIG. 1 is a diagram illustrating cross-correlation characteristics between ZC sequences in combinations of different sequence numbers, which are acquired by computer simulation. To be more specific, FIG. 1 illustrates the cross-correlation characteristics between a ZC sequence of a sequence length N=11 and sequence number r=3, and ZC sequences of a sequence length N=23 and sequence numbers r=1 to 6. In FIG. 1, the horizontal axis represents the delay time using the number of symbols, and the vertical axis represents the normalized cross-correlation values, that is, the values dividing the cross-correlation values by N. As shown in FIG. 1, the maximum cross-correlation value is very high with the combination of a ZC sequence of r=3 and N=11 and a ZC sequence of r=6 and N=23, and is about three times higher than the cross-correlation value in the single transmission bandwidth, 1/√N, that is, 1/√11.
FIG. 2 is a diagram illustrating the inter-cell interference of DM-RS in a case where specific combinations of ZC sequences that increase cross-correlation are allocated to adjacent cells. To be more specific, a ZC sequence of r=a and N=11 and a ZC sequence of r=b and N=23 are allocated to cell #A, and a ZC sequence of r=c and N=23 and a ZC sequence of r=d and N=11 are allocated to cell #B. In this case, the combination of the ZC sequence of r=a and N=11 allocated to cell #A and the ZC sequence of r=c and N=23 allocated to cell #B, or the combination of the ZC sequence of r=b and N=23 allocated to cell #A and the ZC sequence of r=d and N−11 allocated to cell #B, increases the inter-cell interference of DM-RS, and, consequently, degrades the accuracy of channel estimation and degrades the data demodulation performance degrades significantly.
To avoid such problems, the ZC sequence allocating method disclosed in Non-Patent Document 1 is used in a cellular radio communication system. To reduce inter-cell interference, Non-Patent Document 1 suggests allocating a combination of ZC sequences of high cross-correlation and different sequence lengths, to a single cell.
FIG. 3 is a diagram illustrating the ZC sequence allocating methods disclosed in Non-Patent Document 1 and Non-Patent Document 2. In FIG. 3, the example shown in FIG. 2 is used. As shown in FIG. 3, a combination of ZC sequences of high cross-correlation, that is, a combination of a ZC sequence of r=a and N−11 and a ZC sequence of r=c and N−23, is allocated to a single cell (cell #A in this case). Also, another combination of ZC sequences of high cross-correlation, that is, a combination of a ZC sequence of r=d and N−11 and a ZC sequence of r=b and N=23, is allocated to a single cell (cell #B in this case). In the single cell, transmission bands are scheduled by one radio base station apparatus, and, consequently, ZC sequences of high correlation allocated to the same cell, are not multiplexed. Therefore, inter-cell interference is reduced.
Also, Non-Patent Document 2 proposes a method of finding a combination of ZC sequence numbers, which are used in RB's (hereinafter referred to as a “sequence group”). ZC sequences have a feature of having higher cross-correlation when the difference of r/N, that is, the difference of sequence number/sequence length is smaller. Therefore, based on a sequence of an arbitrary RB (e.g. one RB), ZC sequences that make the difference of r/N equal to or less than a predetermined threshold, are found from the ZC sequences of each RB, and the multiple ZC sequences found are allocated to a cell as one sequence group.
FIG. 4 is a diagram illustrating a sequence group generation method disclosed in Non-Patent Document 2. In FIG. 4, the horizontal axis represents r/N, and the vertical axis represents the ZC sequence of each RB. First, the reference sequence length Nb and reference sequence number rb are set. Hereinafter, a ZC sequence having the reference sequence length Nb and reference sequence number rb is referred to as a “reference sequence.” For example, if Nb is 13 (which is the sequence length associated with one RB) and rb is 1 (which is selected between 1 and Nb−1), rb/Nb is 1/13. Next, ZC sequences that make the difference of r/N from the reference rb/Nb equal to or less than a predetermined threshold, are found from the ZC sequences of each RB to generate a sequence group. Also, the reference sequence number is changed, and, in the same process as above, other sequence groups are generated. Thus, it is possible to generate different sequence groups for the number of reference sequence numbers, that is, it is possible to generate N b−1 different sequence groups. Here, if ranges for selecting ZC sequences, in which a difference from rb/Nb is equal to or less than a predetermined threshold, overlap between adjacent sequence groups, the same ZC sequences are included in the plurality of sequence groups, and therefore the sequence numbers overlap between cells. Therefore, to prevent ranges for selecting ZC sequences in adjacent sequence groups from overlapping, the above predetermined threshold is set to, for example, a value less than 1/(2Nb).
FIG. 5A and FIG. 5B illustrate examples of sequence groups generated by the sequence group generation method disclosed in Non-Patent Document 2. Here, the sequence length N is set to the prime number that is larger than the maximum possible size of transmission in the transmission bandwidth and that is the closest to this size, and, furthermore, the sequence length N is uniquely determined from the number of RB's. FIG. 5A and FIG. 5B illustrate sequence groups (ZC sequence group 1 and ZC sequence group 2) comprised of ZC sequences that satisfy following equation 3 in a case where the reference sequence length Nb is 13 and the reference sequence number rb is 1 or 2. In equation 3, the threshold Xth is, for example, 1/(2Nb), (i.e. 1/26) to prevent the same sequence from being included in a plurality of sequence groups.|rb/Nb−r/N|≦Xth  (Equation 3)
Thus, according to the sequence allocating methods disclosed in Non-Patent Document 1 and Non-Patent Document 2, a sequence group comprised of ZC sequences that make a difference of r/N equal to or less than a predetermined threshold, that is, a sequence group comprised of ZC sequences having greater cross-correlation than a predetermined threshold, is generated, and the generated sequence group is allocated to the single cell. By this means, it is possible to allocate a combination of ZC sequences of large cross-correlation and different sequence lengths to the single cell, and reduce inter-cell interference.    Non-Patent Document 1: Huawei R1-070367, “Sequence Allocating method for E-UTRA Uplink Reference Signal”, 3GPP TSG RAN WG1Meeting #47bis, Sorrento, Italy 15-19 Jan., 2007    Non-Patent Document 2: LG Electronics, R1-071542, “Binding method for UL RS sequence with different lengths”, 3GPP TSG RAN WG1Meeting #48bis, St. Julians, Malta, Mar. 26-30, 2007